Google Ads Search Lift: measuring the searches your video drove

Search Lift is the Google Ads study that measures how many extra searches your video campaigns caused. It holds back a control group that doesn't see your ads, compares it with people who did, and reports the increase in searches for terms you choose, on Google and YouTube. It's available for YouTube video campaigns and Demand Gen. Since October 14, 2026, version 26 of the Google Ads API also exposes Search Lift metrics, so you can pull results out of the interface. The main caveat: a search you caused isn't a sale you caused. It's a mid-funnel signal and should be read as one.

What Search Lift measures

Video rarely gets the click. Someone sees your ad on YouTube, closes it, and two days later searches your brand on Google. Last-click gives all the credit to the search, and the video looks like spend with no return. That blind spot is covered in last-click attribution.

Search Lift is built to measure that bridge. How it works, per Google Ads Help:

• Holdout group. Part of the audience doesn't see the ads in the study.
• Study terms. You set up groups of 1 to 3 search terms, usually your brand and product.
• Comparison. Google measures the difference in searches for those terms between people who saw the ads and people who didn't.
• Duration. Studies run 28 days, and results can come in earlier with enough volume.

It doesn't require Brand Lift. You can run Search Lift alone, Brand Lift alone, or both.

When it's worth running

Google says Search Lift budget minimums are similar to Brand Lift, with specifics from your account team. One recommendation straight from Google's help center: include every eligible campaign in the study. Leave one out and the control group still sees ads, which dilutes the effect.

How to read the results

A study comes back in one of three states:

• X% relative lift. There was enough data, and the lift is statistically significant.
• Not enough data. Impressions or searches fell below the threshold. That's not "it didn't work", it's "it couldn't be measured".
• No lift detected. There was enough data, but no significant difference between the groups.

And the metrics, in the order worth reading them:

1. Relative lift. The absolute increase in searches divided by baseline searches. It's the headline number.
2. Certainty of lift. Calculated as 1 minus the p-value: the higher it is, the less likely the result is chance. Read it with the confidence interval, which narrows as data accumulates.
3. Incremental searches per impression and per cost. These are indexes: above 1, that slice performed better than the study average. Use them to compare campaigns, formats and audiences.
4. By search property and by term. How much of the lift came from Google versus YouTube, and which terms drove it.

A worked example: say the study shows a 15% relative lift in brand searches with high certainty, and the per-cost index is 1.4 for Shorts versus 0.8 for in-stream. The honest conclusion is "the video drove brand searches, and Shorts drove more per dollar". It's not yet "the video drove sales".

The common mistake: treating a search as a conversion

Search Lift answers "did the ads make more people search?". It doesn't answer:

• Whether the search turned into a click. They may have searched and clicked a competitor.
• Whether the click turned into a lead. That depends on the landing page and the offer.
• Whether the lead turned into a sale. That depends on sales, and only the CRM knows.
• Whether the extra search was expensive. Some of it may have been paid for again by your branded Search campaign.

The full picture takes three readings stacked on top of each other: Search Lift says searches went up, your Search campaigns and Search Console say how many of those searches became clicks, and the CRM says how many became sales. For a causal read on sales, you need a different design: a holdout test measuring conversions, as laid out in incrementality testing.

What changes with API v26

Until now, Search Lift results lived in the Google Ads interface. With version 26 of the API, released October 14, 2026, Search Lift metrics (such as relative lift and incremental searches) come through the API, segmented by the study's keyword groups and by search property.

In practice, that means you can put results in your own report next to spend and conversions, without copying from the screen. For agencies, agree with the client before the study which terms go in and what lift would count as good, so the number doesn't become an argument later. If brand presence in search is the broader question, share of voice adds context.

Where Vazante fits

Next step

If you run YouTube or Demand Gen and nobody can say what the video does for the brand, ask your account team for a Search Lift study, define the terms up front and include every eligible campaign. While it runs, set up the next two readings: brand search in Search Console and sales in your CRM. To see Google Ads, Meta and TikTok against the same yardstick in the meantime, see paid ads reporting. If you're weighing YouTube against other video channels, TikTok Ads vs YouTube Ads is a useful companion.